The accuracy of functional parameters extracted from ventricular time-activity curves by Fourier curve fitting: a simulation study.

The accuracy of functional parameters derived from multiple harmonic Fourier curve fitting of left ventricular time-activity curves (TACs) was assessed in simulated curves of exactly known characteristics. Curves were generated to represent TACs of equilibrium gated radionuclide ventriculography (ERNV) sampling at 16 frames cycle-1. Curves of varying shape and noise characteristics were studied. The number of harmonics used in the fitting process was variable, the number being determined by an automated 'goodness of fit' test based on the chi-squared test. The accuracy of measured peak ejection rate (PER) and peak filling rate (PFR) was independent of the shape of the curve, but the accuracy of the time to these slopes (TPER, TPFR) was dependent on the steepness of the slope. The accuracy of the measured parameters improved as the influence of noise decreased allowing more harmonics to be added. The values of PER and PFR in control subjects varied between individuals, and with heart rate. There was less variation in the ratio (PER/PFR) of the slopes between individuals, and the accuracy of the ratio was better than that of the individual values of PER and PFR. Functional parameters of predictable accuracy may be determined by Fourier curve fitting at the statistical quality of global or regional ventricular TACs derived from ERNV sampling at 16 frames cycle-1.